KR20220122715A - Computerized systems and methods for predicting and managing scrap - Google Patents

Abstract

A system and method are disclosed for predicting the likelihood that an inventory holding unit (SKU) will be scrapped and taking preventive action accordingly. The system receives a request to initiate a process associated with predicting scrap likelihood associated with a plurality of SKUs, inputs data associated with each SKU into a trained predictive model, and generates predictions for each SKU from the trained predictive model. It outputs, computes the set of values to measure the trained predictive model, and creates a matrix containing the set of computed values. Critical predictions are determined using the matrix, and each SKU prediction is compared to a threshold prediction. Based on the comparison, it is predicted that any one of the plurality of SKUs is likely to be scrapped and at least one of the plurality of preventive measures is performed.

Description

Computerized systems and methods for predicting and managing scrap

CROSS-REFERENCE TO RELATED APPLICATIONS

This application seeks the benefit of priority from U.S. Provisional Application No. 63/123,642, filed on December 10, 2020. The disclosure of this application is expressly incorporated herein by reference in its entirety.

technical field

The present disclosure relates generally to computerized systems and methods for predicting and preventing scrap. In particular, embodiments of the present disclosure determine whether one or more items with a particular stock keeping unit (SKU) are likely to be scrapped in the near future by using a trained neural network to compute predictive values, and It relates to a unique and non-traditional system for preventing an item from being scraped by taking preventive action when it is determined based on a threshold that an item is likely to be scrapped.

Fresh food inventory management presents many challenges to almost any e-commerce company. When placing a purchase order (PO), there is a target demand period from the expected delivery date to the end of the demand period. After the demand period, if the inventory exceeds its sell-by date, it must be discarded. Inventory that must be disposed of, also known as scrap, presents a key challenge to making fresh food retail profitable. To prevent scrap, order quantities can be limited to a lower level, but doing so will result in inventory issues and loss of gross merchandise value (GMV). Therefore, scrap disposal is not a problem that can be solved simply by lowering the inventory ratio.

To alleviate this problem, traditional electronic prediction systems implement expert systems based on manually designed rules or use decision tree models. For example, an expert system can be built by creating a rule database that determines whether an article is expected to be scrapped in the near future. Such a system attempts to accurately predict scrap, but the number of factors affecting scrap is very large, so it is not suitable for making it too complicated to establish a clear rule for judging the risk of scrap. Predicting scrap requires being able to collect as many relevant factors as possible without having to worry about noise due to unrelated factors.

Accordingly, there is a need for an improved method and system for predicting that an article will be scrapped by implementing a predictive model that can accept various kinds of factors as input.

One aspect of the present disclosure relates to a system comprising one or more memory devices storing instructions and one or more processors configured to execute instructions to perform a method for determining scrapability of an inventory holding unit. The method includes receiving a request to initiate a process associated with predicting scrap potential associated with a plurality of inventory holding units (SKUs), responsive to receiving the request, preparing input data associated with each SKU of a plurality of SKUs and - inputting input data associated with each SKU of the plurality of SKUs into the trained predictive model, outputting a predicted value for each SKU of the plurality of SKUs from the trained predictive model, calculating a set of values preparing the information associated with the request by - measuring the trained predictive model by ; and generating a matrix comprising at least a subset of the set of computed values. The method includes selecting a first value of a set of calculated values corresponding to a threshold prediction value of a plurality of thresholds, comparing a prediction value output from a trained prediction model for each SKU among a plurality of SKUs with a threshold prediction value determining a scrap probability for each SKU of the plurality of SKUs based on the comparison, and performing at least one of the plurality of preventive actions based on the scrap probability indicating that scrap is likely. include more

Another aspect of the present disclosure relates to a method for determining the scrapability of an inventory holding unit. The method includes receiving a request to initiate a process associated with predicting scrap potential associated with a plurality of inventory holding units (SKUs), responsive to receiving the request, preparing input data associated with each SKU of a plurality of SKUs and - inputting input data associated with each SKU of the plurality of SKUs into the trained predictive model, outputting a predicted value for each SKU of the plurality of SKUs from the trained predictive model, calculating a set of values preparing the information associated with the request by - measuring the trained predictive model by ; and generating a matrix comprising at least a subset of the set of computed values. The method includes selecting a first value of a set of calculated values corresponding to a threshold prediction value of a plurality of thresholds, comparing a prediction value output from a trained prediction model for each SKU among a plurality of SKUs with a threshold prediction value determining a scrap probability for each SKU of the plurality of SKUs based on the comparison, and performing at least one of the plurality of preventive actions based on the scrap probability indicating that scrap is likely. include more

Another aspect of the present disclosure relates to a system comprising one or more memory devices storing instructions and one or more processors configured to execute instructions to perform a method for determining scrapability of an inventory holding unit. The method includes receiving a request to initiate a process associated with predicting scrap likelihood associated with a plurality of inventory holding units (SKUs), and in response to receiving the request, input data associated with each SKU of a plurality of SKUs. preparing, and - inputting input data associated with each SKU of the plurality of SKUs into the trained neural network, outputting a predicted value for each SKU of the plurality of SKUs from the trained neural network, calculating a set of values preparing information associated with the request by measuring the trained neural network, and generating a matrix comprising at least a subset of the set of computed values. The method includes selecting a first value of a set of calculated values corresponding to a threshold prediction value of a plurality of threshold values, and comparing a prediction value output from a trained neural network for each SKU among the plurality of SKUs with a threshold prediction value , determining a scrap probability for each SKU of the plurality of SKUs based on the comparison, and performing at least one of the plurality of preventive actions based on the scrap probability indicating that scrap is likely. include

Other systems, methods, and computer-readable media are also discussed herein.

1A is a schematic block diagram illustrating an exemplary embodiment of a network including a computerized system for communication that enables transportation, transport, and logistic operations consistent with the disclosed embodiments.
1B illustrates a sample Search Result Page (SRP) that includes one or more search results satisfying a search request, in conjunction with interactive user interface elements, consistent with disclosed embodiments.
1C illustrates a sample Single Detail Page (SDP) containing information about a product and product, along with interactive user interface elements, consistent with disclosed embodiments.
1D illustrates a sample cart page that includes items into a virtual shopping cart, along with interactive user interface elements, consistent with disclosed embodiments.
1E illustrates a sample order page including items from a virtual shopping cart along with information regarding purchases and shipping, along with interactive user interface elements, consistent with the disclosed embodiments.
2 is a schematic diagram of an exemplary fulfillment center configured to utilize the disclosed computerized system, consistent with the disclosed embodiments.
3A and 3B illustrate example methods for predicting scrap likelihood of an inventory holding unit and preventing scrap, consistent with the disclosed embodiments.
3C illustrates an exemplary method for preparing input data for a predictive model, consistent with the disclosed embodiments.
4 depicts a block diagram illustrating an exemplary flow of preparing input data for a predictive model, consistent with the disclosed embodiments.
5 depicts a diagram illustrating an exemplary flow of information through a predictive model, consistent with the disclosed embodiments.
6 depicts a table representing an exemplary generated matrix, consistent with the disclosed embodiments.

The following detailed description refers to the accompanying drawings. Wherever possible, the same reference numbers are used in the drawings and the following description to refer to the same or like parts. While several exemplary embodiments have been described herein, modifications, adaptations, and other implementations are possible. For example, substitutions, additions, or modifications may be made to the components and steps shown in the figures, and the exemplary methods described herein may be modified by substituting, rearranging, removing, or adding steps in the disclosed methods. can Accordingly, the detailed description that follows is not limited to the disclosed embodiments and examples. Instead, the proper scope of the invention is defined by the appended claims.

An embodiment of the present disclosure executes data through a trained neural network, measures the trained neural network by calculating a set of values to determine a threshold prediction value, and combines the data output by the trained neural network with the threshold prediction value. A system and method configured to determine whether one or more items having a particular SKU (stock holding unit) are likely to be scrapped in the near future by comparing and determining whether the item is likely to be scrapped. A preventive action may be taken based on a determination that an article having an inventory holding unit is likely to be scrapped. The disclosed embodiment provides an innovative technical feature that allows for proactive management by predicting that an item will be scrapped, the prediction being achieved by running a trained neural network and calculating a set of values to measure the performance of the neural network. For example, the disclosed embodiment makes it possible to input a plurality of scaled factors into a trained neural network to output predictive values for a plurality of SKUs, and to calculate a set of values to measure the trained neural network by calculating a set of values. enable, and enable determination of threshold predictions for comparison with predictions output by the trained neural network, enable determination of whether an SKU is likely to be scrapped, and when it is determined that an SKU is likely to be scrapped. make it possible to carry out preventive measures;

1A , shown is a schematic block diagram 100 illustrating an exemplary embodiment of a system including a computerized system for communications that facilitates transport, transport, and logistic operations. As shown in FIG. 1A , system 100 may include a variety of systems, each of which may be connected to each other via one or more networks. The systems may also be connected to each other via a direct connection using, for example, a cable. The systems shown include a shipping authority technology (SAT) system 101 , an external front end system 103 , an internal front end system 105 , and a transportation system. (transportation system) 107, mobile devices 107A, 107B and 107C, merchant portal 109, shipment and order tracking (SOT) system 111, fulfillment optimization (FO) system 113 , fulfillment messaging gateway (FMG) 115 , supply chain management (SCM) system 117 , warehouse management system 119 , mobile device 119A , 119B and 119C (shown as being internal to a fulfillment center (FC) 200), third party fulfillment systems 121A, 121B and 121C, a fulfillment center authorization system (FC). Auth) 123 and a labor management system (LMS) 125 .

In some embodiments, the SAT system 101 may be implemented as a computer system that monitors order status and delivery status. For example, the SAT system 101 may determine whether an order has passed its Promised Delivery Date (PDD), initiate a new order, reship items in an undelivered order, Appropriate actions may be taken, including canceling undelivered orders, initiating contact with the customer who placed the order, and the like. The SAT system 101 also provides other data, including outputs (such as the number of packages shipped during a specified period) and inputs (such as the number of empty cardboard boxes received for use in shipment). can be monitored. SAT system 101 also acts as a gateway between different devices in system 100 , such as an external front end system (eg, using store-and-forward or other techniques). Enables communication between devices such as 103 and FO system 113 .

In some embodiments, external front end system 103 may be implemented as a computer system that allows external users to interact with one or more systems within system 100 . For example, in an embodiment where the system 100 enables presentation to the systems to allow a user to order an item, the external front end system 103 receives the search request, presents the item page, and , it can be implemented as a web server that solicits payment information. For example, the external front end system 103 may be implemented as a computer or computers running software such as Apache HTTP Server, Microsoft Internet Information Services (IIS), NGINX, and the like. In another embodiment, the external front end system 103 executes customer web server software designed to receive and process requests from external devices (eg, mobile device 102A or computer 102B), and respond to these requests. may obtain information from databases and other data stores based on the information, and provide a response to a received request based on the obtained information.

In some embodiments, the external front end system 103 may include one or more of a web caching system, a database, a search system, or a payment system. In one aspect, the external front end system 103 may include one or more of these systems, while in another aspect, the external front end system 103 may include an interface (eg, an interface) coupled to one or more of these systems. server-to-server, database-to-database, or other network connection).

An exemplary set of steps illustrated by FIGS. 1B , 1C , 1D and 1E will help explain some operation of the external front end system 103 . External front end system 103 may receive information from systems or devices within system 100 for presentation and/or display. For example, the external front end system 103 may include a search results page (SRP) (eg, FIG. 1B), a single detail page (SDP) (eg, FIG. 1C), a cart page (eg, FIG. 1C). 1d) or an order page (eg, FIG. 1E ), may host or serve one or more web pages. A user device (eg, using a mobile device 102A or computer 102B) may request a search by searching the external front end system 103 and entering information in a search box. External front end system 103 may request information from one or more systems within system 100 . For example, the external front end system 103 may request information from the FO system 113 that satisfies the search request. The external front-end system 103 may also request and receive (from the FO system 113) the promised delivery date or "PDD" for each product included in the search results. In some embodiments, the PDD determines that within a certain period of time, for example, by the end of the day (11:59 PM), if ordered, when a package containing the product arrives at the user's desired location, or the product is It may represent an estimate for a date promised to be delivered to a user's desired location. (PDD is discussed further below with respect to FO system 113).

The external front end system 103 may prepare an SRP (eg, FIG. 1B ) based on the information. The SRP may contain information that satisfies the search request. For example, it may include pictures of products that satisfy the search request. The SRP may also include information about each price of each product, or enhanced shipping options for each product, PDD, weight, size, offers, discounts, and the like. The external front end system 103 may send the SRP to a requesting user device (eg, via a network).

The user device can then select a product from the SRP by selecting the product represented on the SRP, for example, by clicking or tapping the user interface or using another input device. The user device may formulate a request for information on the selected product and send it to the external front end system 103 . In response, the external front end system 103 may request information associated with the selected product. For example, the information may include additional information other than what is presented for the product on the respective SRP. This may include, for example, shelf life, country of origin, weight, size, number of items in package, handling instructions or other information about the product. may include This information may include recommendations for similar products (e.g., based on big data and/or machine learning analysis of customers who have purchased this product and at least one other product), often It may include answers to questions asked, reviews from customers, manufacturer information, photos, and the like.

The external front-end system 103 may prepare a Single Detail Page (SDP) (eg, FIG. 1C ) based on the received product information. The SDP may also include other interactive elements such as a “Buy Now” button, an “Add to Cart” button, a quantity field, a picture of the item, and the like. The SDP may further include a list of sellers offering products. The list may be ordered based on the price offered by each seller, so that the seller offering to sell the product at the lowest price may be listed at the top. The list may also be ordered based on the seller ranking such that the highest ranked seller may be listed at the top. Merchant rankings may be formulated based on a number of factors, including, for example, historical tracking records of sellers who have met their promised PDDs. The external front end system 103 may forward the SDP to the requesting user device (eg, via a network).

The requesting user device may receive an SDP listing product information. Upon receiving the SDP, the user device may interact with the SDP. For example, the user of the requesting user device may click or otherwise interact with a "Place in Cart" button on the SDP. This adds the product to a shopping cart associated with the user. The user device may send this request to the external front end system 103 to add the product to the shopping cart.

The external front end system 103 may create a cart page (eg, FIG. 1D ). In some embodiments, the cart page lists products that the user has added to a virtual "shopping cart." A user device may request a cart page by clicking or otherwise interacting with an icon on the SRP, SDP or other pages. In some embodiments, the cart page lists all products that the user has added to the shopping cart, as well as the quantity of each product, the price per item of each product, the price based on the associated quantity of each product, information about the PDD; User interface elements for modifying shipping methods, shipping costs, products in the shopping cart (eg, deleting or modifying quantities), options for ordering other products or setting up periodic delivery of products You can list information about the products in your cart, such as items, options for setting up an interest payment, user interface elements for making a purchase, and the like. A user at the user device may click or otherwise interact with a user interface element (eg, a button that reads “Buy Now”) to initiate a purchase of a product in the shopping cart. In doing so, the user device may send this request to the external front end system 103 to initiate a purchase.

The external front end system 103 may generate an order page (eg, FIG. 1E ) in response to receiving a request to initiate a purchase. In some embodiments, the order page re-lists items from the shopping cart and requests entry of payment and shipping information. For example, an order page may contain a section requesting information about a buyer of items in a shopping cart (eg, name, address, email address, phone number), information about a recipient (eg, name, address, phone number, shipping information), shipping information (eg speed/method of delivery and/or pickup), payment information (eg credit card, bank transfer, check, stored credit) ), a user interface element for requesting a cash receipt (eg, for tax purposes), etc. The external front end system 103 may send an order page to the user device.

The user device may click or otherwise interact with a user interface element that enters information on the order page and transmits the information to the external front end system 103 . From there, the external front end system 103 may transmit information to different systems within the system 100 to enable the creation and processing of new orders with products in the shopping cart.

In some embodiments, the external front end system 103 may be further configured to enable merchants to send and receive information regarding orders.

In some embodiments, the internal front end system 105 allows internal users (eg, employees of an organization that owns, operates, or leases the system 100 ) to interact with one or more systems within the system 100 . It can be implemented as a computer system that allows For example, in an embodiment where system 100 enables presentation of systems that allow users to order items, internal front end system 105 allows internal users to view diagnostic and statistical information about orders, or It can be implemented as a web server that allows to view, edit product information, or review statistics about orders. For example, the internal front end system 105 may be implemented as a computer or computers running software such as Apache HTTP Server, Microsoft Internet Information Services (IIS), NGINX, and the like. In another embodiment, the internal front end system 105 may run customer web server software designed to receive and process requests from the systems or devices depicted in system 100 (as well as other devices not shown). and obtain information from databases and other data stores based on these requests, and provide a response to the received request based on the obtained information.

In some embodiments, the internal front end system 105 may include one or more of a web caching system, a database, a search system, a payment system, an analytics system, an order monitoring system, and the like. In one aspect, the internal front end system 105 may include one or more of these systems, while in another aspect, the internal front end system 105 may include interfaces (e.g., , server-to-server, database-to-database, or other network connections).

In some embodiments, transportation system 107 may be implemented as a computer system that enables communication between systems or devices within system 100 and mobile devices 107A-107C. In some embodiments, transportation system 107 may receive information from one or more mobile devices 107A - 107C (eg, mobile phones, smart phones, PDAs, etc.). For example, in some embodiments, mobile devices 107A-107C may include devices operated by delivery workers. Delivery workers, which may be permanent, temporary, or shift workers, may utilize mobile devices 107A-107C to effect delivery of packages containing products ordered by users. . For example, to ship a package, the delivery operator may receive a notification on the mobile device indicating which package to ship and where to ship the package. Upon arrival at the delivery location, the delivery operator places the package (eg, in the back of a truck, or in a crate of packages) or uses a mobile device to identify an identifier on the package (eg, By scanning or otherwise capturing data associated with barcodes, images, text strings, RFID tags, etc., and by placing packages (e.g., at your doorstep, with security guards, or handing over to recipients, etc.) ) can be transmitted. In some embodiments, the delivery worker may use a mobile device to capture a photo(s) of the package and/or obtain a signature. The mobile device sends information including information about the shipment to the transportation system 107 including, for example, time, date, GPS location, photo(s), an identifier associated with the delivery operator, an identifier associated with the mobile device, and the like. can be transmitted Transportation system 107 may store this information in a database (not shown) for access by other systems in system 100 . In some embodiments, the transportation system 107 may use this information to prepare and transmit tracking data indicating the location of a particular package to other systems.

In some embodiments, certain users can use one type of mobile device (eg, full-time workers can use a specialized PDA with custom hardware such as barcode scanners, styluses, and other devices). may be used), while different users may use different types of mobile devices (eg, temporary or shift workers may utilize off-the-shelf mobile phones and/or smart phones). .

In some embodiments, transportation system 107 may associate a user with each device. For example, the transportation system 107 may include a user (eg, represented by a user identifier, employee identifier, or phone number) and a mobile device (eg, International Mobile Equipment Identity (IMEI), International Mobile Subscription (IMSI)). Identifier), a phone number, represented as a Universal Unique Identifier (UUID), or a Globally Unique Identifier (GUID)). Transportation system 107 may use this association with data received during shipment to analyze data stored in a database to determine, among other things, the location of the operator, the efficiency of the operator, or the speed of the operator.

In some embodiments, merchant portal 109 may be implemented as a computer system that enables merchants or other external entities to communicate electronically with one or more systems within system 100 . For example, a seller may use the seller portal 109 to upload or provide product information, order information, contact information, etc. for products that the seller wishes to sell through the system 100 to a computer system (shown as shown). not available) can be used.

In some embodiments, the shipping and order tracking system 111 receives information regarding the location of packages containing products ordered by customers (eg, by a user using devices 102A and 102B). It can be implemented as a computer system that stores, stores, and transmits. In some embodiments, the shipping and order tracking system 111 requests or stores information from web servers (not shown) operated by shipping companies that ship packages containing products ordered by customers. can

In some embodiments, the shipping and order tracking system 111 may request and store information from systems depicted within the system 100 . For example, the shipping and order tracking system 111 may request information from the transportation system 107 . As discussed above, transportation system 107 may include one or more mobile devices 107A-107C (eg, 107A-107C) associated with one or more users (eg, delivery workers) or vehicles (eg, delivery trucks). information can be received from a mobile phone, smart phone, PDA, etc.). In some embodiments, the shipping and order tracking system 111 also provides a warehouse management system (WMS) 119 to determine the location of individual products within a fulfillment center (eg, fulfillment center 200 ). ) can be requested from Shipment and order tracking system 111 requests data from one or more of transportation system 107 or WMS 119 , processes the data, and sends the data upon request to devices (eg, user devices 102A and 102B). )) can be presented.

In some embodiments, the fulfillment optimization (FO) system 113 is configured to respond to customer orders from other systems (eg, external front end system 103 and/or shipping and order tracking system 111 ). It can be implemented as a computer system that stores information about the The FO system 113 may also store information describing where particular items are held or stored. For example, certain items may be stored in only one fulfillment center, while certain other items may be stored in multiple fulfillment centers. In another embodiment, certain fulfillment centers may be designed to store only a certain set of items (eg, fresh or frozen products). The FO system 113 stores this information as well as related information (eg, quantity, size, date of receipt, expiration date, etc.).

The FO system 113 may also calculate a corresponding promised delivery date (PDD) for each product. In some embodiments, PDD may be based on one or more factors. For example, the FO system 113 may determine the historical demand for the product (eg, how often the product was ordered during a period), the expected demand for the product (how many customers will order the product in the coming period), and forecast), network-wide historical demand indicating how many products were ordered during a period, network-wide projected demand indicating how many products are expected to be ordered in the coming period, and each fulfillment center For the products stored in 200 , the PDD for the product may be calculated based on one or more counts stored by the fulfillment center for each product, an expected order or a current order for the product, and the like.

In some embodiments, the FO system 113 periodically (eg, hourly) determines the PDD for each product and stores it in a database for retrieval, or other systems (eg, For example, external front-end system 103 , SAT system 101 , shipment and order tracking system 111 ). In another embodiment, the FO system 113 may send electronic requests from one or more systems (eg, external front end system 103 , SAT system 101 , shipment and order tracking system 111 ). ) and can compute the PDD on demand.

In some embodiments, a fulfillment messaging gateway (FMG) 115 receives a request or response in one format or protocol from one or more systems in system 100 , eg, FO system 113 , and or converting the response to another format or protocol, and forwarding the request or response in the converted format or protocol to another system, for example, the WMS 119 or a third-party fulfillment system 121A, 121B or 121C, or , and vice versa.

In some embodiments, the supply chain management (SCM) system 117 may be implemented as a computer system that performs predictive functions. For example, the SCM system 117 may include, for example, historical demand for products, predicted demand for products, network-wide historical demand, network-wide predicted demand, products stored in each fulfillment center 200 , for example. A level of demand for a specific product may be predicted based on a field count, an expected order or a current order for each product, and the like. In response to this predicted level and quantity for each product across all fulfillment centers, the SCM system 117 sets up one or more purchase orders to purchase and stock up sufficient quantities to satisfy the predicted demand for that particular product. can create

In some embodiments, warehouse management system (WMS) 119 may be implemented as a computer system that monitors the workflow. For example, WMS 119 may receive event data from a separate device (eg, devices 107A-107C or 119A-119C) representing discrete events. For example, WMS 119 may receive event data indicating the use of one of these devices to scan a package. As discussed below with respect to the fulfillment center 200 and FIG. 2 , during the fulfillment process, a package identifier (eg, barcode or RFID tag data) is transferred to machines (eg, automated or devices such as a portable barcode scanner, RFID reader, high-speed camera, tablet 119A, mobile device/PDA 119B, computer 119C, etc.). WMS 119 may store each event representing a scan or read of a package identifier in a corresponding database (not shown) along with package identifier, time, date, location, user identifier or other information, and store this information in other system (eg, shipping and order tracking system 111 ).

In some embodiments, WMS 119 may store information relating one or more devices (eg, devices 107A-107C or 119A- 119C) with one or more users associated with system 100 . For example, in some circumstances, a user (eg, a part-time or full-time employee) may be associated with a mobile device in that the user owns the mobile device (eg, the mobile device is a smart phone). In another situation, the user may be aware that the user is temporarily in possession of the mobile device (eg, the user checks out the mobile device at the start of the day, will use it for the day, and return it at the end of the day) , may be associated with a mobile device.

In some embodiments, WMS 119 may maintain a work log for each user associated with system 100 . For example, WMS 119 may perform any assigned process (eg, unloading trucks, picking items from a pick zone, rebin wall work, packing items). below), user identifier, location (e.g., floor or area within fulfillment center 200), number of units moved through the system by staff (e.g., number of picked items, number of packaged items), number), an identifier associated with a device (eg, devices 119A- 119C), and the like. In some embodiments, WMS 119 may receive check-in and check-out information from a timekeeping system, such as a timekeeping system operating on devices 119A-119C.

In some embodiments, third-party fulfillment (3PL) systems 121A-121C represent computer systems associated with third-party providers of logistics and products. For example, some products may be stored in the fulfillment center 200 (as described below with respect to FIG. 2 ), while others may be stored off-site or produced on demand; Otherwise stored in the fulfillment center 200 may not be available. 3PL systems 121A - 121C may be configured to receive orders from FO system 113 (eg, via FMG 115 ), products and/or services (eg, delivery or installation) can be provided directly to customers. In some embodiments, one or more of 3PL systems 121A-121C may be part of system 100 , while in other embodiments one or more of 3PL systems 121A-121C will be external to system 100 . may (eg, owned or operated by a third party provider).

In some embodiments, the Fulfillment Center Authentication System (FC Auth) 123 may be implemented as a computer system having various functions. For example, in some embodiments, FC Auth 123 may act as a single-sign on (SSO) service for one or more other systems within system 100 . For example, FC Auth 123 allows a user to log in through the internal front end system 105 , and similar privileges for the user to access resources in the shipping and order tracking system 111 . ) and allow the user to access these privileges without requiring a second log in process. In another embodiment, FC Auth 123 allows users (eg, employees) to associate themselves with a particular task. For example, some employees may not have electronic devices (eg, devices 119A- 119C), but instead, during the course of a day, within fulfillment center 200 . , by task, and by area, it can be moved. FC Auth 123 can be configured to indicate what tasks these employees are doing and in which zone they are at different times of the day.

In some embodiments, labor management system (LMS) 125 may be implemented as a computer system that stores attendance and overtime information for employees (including full-time and part-time employees). For example, LMS 125 may receive information from FC Auth 123 , WMS 119 , devices 119A- 119C , transportation system 107 , and/or devices 107A- 107C.

The specific configuration shown in FIG. 1A is only an example. For example, while FIG. 1A shows an FC Auth system 123 coupled to an FO system 113 , not all embodiments require this particular configuration. Indeed, in some embodiments, the systems within system 100 are the Internet, Intranet, Wide-Area Network (WAN), Metropolitan-Area Network (MAN), wireless network compatible with IEEE 802.11a/b/g/n standard. , may be connected to each other through one or more public or private networks including a leased line, and the like. In some embodiments, one or more of the systems in system 100 may be implemented as one or more virtual servers implemented in a data center, server farm, or the like.

2 shows a fulfillment center 200 . Fulfillment center 200 is an example of a physical location that stores items for shipping to customers upon order. The fulfillment center (FC) 200 may be divided into a number of zones, each of which is illustrated in FIG. 2 . In some embodiments, these “zones” may be considered virtual divisions between different stages of the process of receiving items, storing items, retrieving items, and transporting items. . Thus, although “zones” are shown in FIG. 2 , other divisions into zones are possible and, in some embodiments, zones in FIG. 2 may be omitted, duplicated, or modified.

Inbound zone 203 represents the area of FC 200 from FIG. 1A where items are received from vendors wishing to sell products using system 100 . For example, the seller may use the truck 201 to deliver the items 202A and 202B. Item 202A may represent a single item large enough to occupy its own shipping pallet, whereas item 202B may represent a set of items stacked together on the same pallet to save space. can indicate

An operator may receive the items in the inbound area 203 and optionally check the items for damage and accuracy using a computer system (not shown). For example, an operator may use a computer system to compare the quantity of items 202A and 202B to an ordered quantity of items. If the quantities do not match, the operator may reject one or more of the items 202A or 202B. Once the quantities match, the operator can move these items (eg, using a dolly, hand truck, forklift, or manually) to the buffer area 205 . Buffer area 205 may be a temporary storage area for items that are not currently needed in the picking area, for example, because there is a sufficient quantity of that item in the picking area to meet the forecasted demand. have. In some embodiments, the forklift 206 is operated to move items around the buffer zone 205 and between the inbound zone 203 and the drop zone 207 . If the item 202A or 202B is needed at the picking area (eg, due to forecasted demand), the forklift may move the item 202A or 202B to the drop area 207 .

Drop zone 207 may be an area of FC 200 that stores articles before they are moved to picking zone 209 . An operator assigned to a picking operation (“picker”) accesses items 202A and 202B at the picking area, scans a barcode for the picking area, and uses a mobile device (e.g., device 119B)). can be used to scan barcodes associated with articles 202A and 202B. The picker may then take the item to the picking area 209 (eg, by carting or transporting the item).

Picking zone 209 may be an area of FC 200 where articles 208 are stored on storage unit 210 . In some embodiments, storage unit 210 may include one or more of physical shelves, bookcases, boxes, totes, refrigerators, freezers, cold stores, and the like. In some embodiments, the picking zone 209 may be comprised of multiple layers. In some embodiments, an operator or machine may manually or in several ways, including, for example, forklifts, elevators, conveyor belts, carts, hand trucks, dollies, automated robots or devices, or devices in picking zone 209 . can be moved to For example, a picker may load items 202A and 202B on hand trucks or carts at drop zone 207 and transport articles 202A and 202B on foot to picking zone 209 .

The picker may receive instructions to place (or "store") the item at particular spots in the picking area 209 , such as a particular space on the storage unit 210 . For example, the picker may scan the article 202A using a mobile device (eg, device 119B). The device may indicate where the picker should receive the item 202A using, for example, an aisle, a shelf, and a system for indicating a position. The device may then prompt the picker to scan the barcode at that location before receiving the item 202A at that location. The device transmits data indicating that the item 202A has been received at the location by the user using the device 119B to a computer system, such as the WMS 119 of FIG. 1A (eg, via a wireless network). can

When a user places an order, the picker may receive a command on device 119B to retrieve one or more items 208 from storage unit 210 . The picker may retrieve the item 208 , scan a barcode on the item 208 , and place the item on a transport mechanism 214 . Although the transport mechanism 214 is shown as a slide, in some embodiments, the transport mechanism may be implemented as one or more of a conveyor belt, elevator, cart, forklift, hand truck, dolly, and the like. Thereafter, the article 208 may arrive at a packing zone 211 .

Packing zone 211 may be an area of FC 200 where goods are received from picking zone 209 and packed into boxes or bags for final shipment to a customer. At the packing area 211 , the worker assigned to receiving the item (“living worker”) will receive the item 208 from the picking area 209 and determine which order it corresponds to. For example, a living operator may scan a barcode on item 208 using a device such as computer 119C. Computer 119C may visually indicate which order the item 208 is associated with. This may include, for example, a “cell” of a space or wall 216 corresponding to an order. Once an order has been placed (eg, because a cell contains all of the items for the order), the live worker can indicate to the packing worker (or "packer") that the order is complete. The packer may retrieve items from the cell and place them in boxes or bags for transport. The packer may then manually or otherwise transport the box or bag to the hub area 213 , for example via a forklift, cart, dolly, hand truck, conveyor belt.

Hub zone 213 may be an area of FC 200 that receives all boxes or bags (“packages”) from packaging zone 211 . Operators and/or machines in the hub area 213 may retrieve the packages 218 , determine which part of the delivery area each package is intended to go to, and route the packages to the appropriate camp area 215 . . For example, if the shipping area has two smaller sub-areas, the package will go to one of the two camp areas 215 . In some embodiments, an operator or machine may scan the package (eg, using one of devices 119A- 119C) to determine its final destination. Routing the package to the camp area 215 includes, for example, determining the portion of the geographic area for which the package is destined (eg, based on a zip code), and a camp area associated with the portion of the geographic area. and determining (215).

In some embodiments, camp area 215 may include one or more buildings, one or more physical spaces or one or more areas, and packages are received from hub area 213 to be classified into routes and/or sub-routes. . In some embodiments, camp area 215 is physically separate from FC 200 , while in other embodiments camp area 215 may form part of FC 200 .

Workers and/or machines within camp area 215 may, for example, compare the destination to existing routes and/or sub-routes, calculate the workload for each route and/or sub-route; Based on the time of day, the method of transportation, the cost to ship the package 220 , the PDD associated with the items in the package 220 , and the like, the package 220 is routed and/or sub-routed along which routes and/or sub-routes. You can decide if it should be related. In some embodiments, an operator or machine may scan the package (eg, using one of devices 119A- 119C) to determine its final destination. Once the packages 220 are assigned to a particular route and/or sub-route, workers and/or machines can move the packages 220 to be transported. In exemplary FIG. 2 , camp area 215 includes truck 222 , automobile 226 , and delivery workers 224A and 224B. In some embodiments, the truck 222 may be driven by a delivery worker 224A, the delivery worker 224A being a full-time employee delivering packages for the FC 200, and the truck 222 being the FC ( 200) is owned, leased, or operated by the same company that owns, leases, or operates. In some embodiments, automobile 226 may be driven by a delivery operator 224B, who delivers on an as-needed basis (eg, seasonally). A "flex" or occasional worker. Vehicle 226 may be owned, leased, or operated by delivery worker 224B.

According to one aspect of the present disclosure, a computer implemented system for predicting scrap likelihood of an inventory holding unit (SKU) includes one or more memory devices to store instructions, and one or more processors configured to execute the instructions to perform an operation. may include The scrap probability of one or more articles with a particular SKU can be predicted by running the data through a deep neural network and measuring the deep neural network to determine a threshold for the data output by the deep neural network. In some embodiments, the disclosed functions and systems may be implemented as part of the SCM system 117 . The preferred embodiment involves implementing the disclosed functions and systems on the SCM system 117 , although other implementations will be appreciated by those skilled in the art.

Scrapability, which indicates the likelihood that an article having a particular inventory holding unit (SKU) will be scraped within a period of time (eg, 2 to 4 days, 48 to 96 hours, etc.) may be determined by considering one or more factors associated with the inventory holding unit. can For example, factors include inventory, sales trend, forecast, out of stock (OOS) time, code life sellable days, average sales, minimum order quantity (MOQ) , shelf life, degree (ie, temperature), condition, lead time, expiration date, exception time, sales delay, and recent scrap. Table 1 is a table of exemplary factors.

index name Justice One shelflife How many days will the SKU product(s) be available for sale? 2 shelflife_3 shelflife < 3 days 3 shelflife_7 shelflife < 7 days 4 shelflife_14 shelflife < 14 days 5 shelflife_28 shelflife < 28 days 6 shelflife_181 shelflife < 181 days 7 degree_chilled Whether the SKU product(s) should be refrigerated 8 degree_frozen Whether the SKU product(s) should be frozen 9 degree_ambient Whether the SKU product(s) should be stored at ambient temperature 10 new_sku Whether the SKU is new (i.e. first seen by the system within the last 6 months) 11 obselecte_sku 'OBSOLETE' in postatus; Indicates whether a SKU is obsolete (i.e. deprecated for that product) 12 inactive_sku 'INACTIVE_SEASON_OFF' in postatus; Indicates whether the SKU is inactive (e.g. winter jacket during summer) 13 manual_sku 'MANUAL' in postatus; Manually enter the purchase order status for the SKU if the SKU does not belong to one of the above statuses 14 leadtime number of days between order and delivery 15 leadtime_2 leadtime < 2 days 16 leadtime_3 leadtime < 3 days 17 leadtime_4 leadtime < 4 days 18 leadtime_10 leadtime < 10 days 19 leadtime_30 leadtime < 30 days 20 leadtime_90 leadtime < 90 days 21 expirationdate expire_date_managed = 'Y' (i.e. the expiration date of the SKU) 22 dp Demand period days (i.e. a value representing the target sales period of the SKU) 23 dp_2 dp < 2 days 24 dp_3 dp < 3 days 25 dp_4 dp < 4 days 26 dp_7 dp < 7 days 27 dp_14 dp < 14 days 28 moq Minimum order quantity (e.g. minimum number of units purchased at one time) 29 sellable_days Number of days sold indicating whether the SKU has a short code life (i.e. high order frequency) or a long code life (low order frequency) 30 sellable_days_3 sellable_days < 3 days 31 sellable_days_4 sellable_days < 4 days 32 sellable_days_5 sellable_days < 5 days 33 sellable_days_7 sellable_days < 7 days 34 sellable_days_14 sellable_days < 14 days 35 sellable_days_28 sellable_days < 28 days 36 sellable_days_181 sellable_days < 181 days 37 exp_sales_dp Total estimated daily sales during the day of the demand period 38 inv Inventory quantity on the day of order 39 avg_adj_sales Average adjusted sales over last 7 days 40 avg_sales Average Actual Sales over Last 7 Days 41 andon_sku Yesterday's andon (i.e. the time yesterday's SKU was unsellable) 42 AVG_D3_D Average sales from D-3 (3 days before D day) to D day 43 MAX_D3_D Maximum sales from D-3 (3 days before D day) to D day 44 MAX_D7_D3 Maximum sales from D-7 (7 days before D day) to D-3 (3 days before D day) 45 AVG_D7_D3 Average sales from D-7 (7 days before D day) to D-3 (3 days before D day) 46 over_promo_times Actual Sales > After Promo Prediction 47 overforecast_2_times Actual number of sales > exp_sales_dp (index 37) * 2 48 over_order_12_times Purchase order quantity > exp_sales_dp (index 37) * 1.2 49 long_sl_overforecast_2_times Shelf life > 14 days 50 long_sl_over_order_12_times Shelf life > 14 days 51 one_day_oos_hours Number of hours SKUs were out of stock (oos) yesterday 52 one_day_andon_hours The number of hours the SKU was unsellable (and not) yesterday. 53 one_day_exclude_sales_hours The number of hours the SKU was excluded from sale yesterday. 54 one_day_discount_hours Number of hours the SKU was discounted yesterday 55 two_days_oos_hours Number of hours the SKU was out of stock in the last 2 days 56 two_days_andon_hours Number of hours the SKU was unsellable (and not) in the last 2 days 57 two_days_exclude_sales_hours The number of hours the SKU was excluded from sale in the last two days. 58 two_days_discount_hours Number of hours the SKU was discounted in the last 2 days 59 sales_delayed_avg Average of sales_dt days after the end of the demand period 60 sales_delayed_avg_sellable sales_delayed_avg > sellable_days 61 sales_delayed_avg_sellable_v2 sellable_days - sales_delayed_avg 62 sales_delayed_dts Count of sales_dt days after the end of the demand period 63 near_scrap_times Number of sales_after_edd_days + 1 >= sellable_days
(edd = expected delivery date) 64 yesterday_sales_after_edd_days Yesterday's sales_after_edd_days
(edd = expected delivery date) 65 stow_delay When the actual receipt completion time is later than the expected delivery completion time of the Fulfillment Center 66 receive_delay If the actual reception time is later than the expected reception time of the fulfillment center 67 target_scrap_inv Sum of target_scrap_inventory within range days 68 sales_1d Yesterday's Actual Sales 69 adj_sales_1d Yesterday's Adjusted Sales 70 sales_2d Actual sales in the last 2 days 71 adj_sales_2d Adjusted Sales for the Last 2 Days 72 sales_3d Actual sales in the last 3 days 73 adj_sales_3d Adjusted Sales for the Last 3 Days 74 po_diff_roq recent average ([purchase_order] - [recommended_order_quantity]) / avg_sales 75 roq_diff_fcst Average of recent ([recommended_order_quantity] - [forecast]) / avg_sales 76 actual_po Average of recent actual purchase_orders / avg_sales 77 xdock_roq_diff_fcst Recommended order quantity different from the prediction for cross docking case 78 xdock_fc_order_qty Fulfillment Center Order Quantity for Cross Docking Cases 79 safety stock safety_stock / 100 (safety stock = additional inventory needed to reduce out of stock) 80 over_fct_times For Promotion Day, Forecast > Actual Sales 81 promo_times Number of promotions in the last 10 days 82 overforecast_1_times Number of forecasts > expected_sales_demand_period 83 overforecast_1_2_times Number of forecasts > expected_sales_demand_period * 1.2 84 over_order_times number of times purchase_order_quantity > expected_sales_demand_period 85 total_scrap_units_recent_1d Yesterday's scrap unit total 86 total_scrap_units_recent_3d Total scrap units for the last 3 days 87 actual_po Overlapping #76 for constructing convolutional neural networks 88 inv_exp_sales_dp inventory - exp_sales_dp 89 moq_sales_sellable moq > avg_sales * sellable_days 90 small_inv inventory < adj_avg_sales 91 sales_drop AVG_D7_D3 > AVG_D3_D * 1.5 && MAX_D7_D3 > MAX_D3_D * 1.5 92 sales_increase AVG_D3_D > AVG_D7_D3 * 1.5 && MAX_D3_D > MAX_D7_D3 * 1.5 93 large_moq moq > avg_adjusted_sales * sellable_days

These factors may be input to a trained neural network, and the trained neural network outputs a predicted value indicating the scrap probability associated with the SKU based on a threshold value. There are some SKUs that are expected to be scrapped depending on the threshold, which can affect the total loss of merchandise value (GMV). In order to maximize the reduced scrap while also minimizing the GMV loss, a threshold such that the two variables are balanced can be determined by generating a matrix containing the calculated values such as recall and precision. Once the threshold is determined, the scrap likelihood associated with the SKUs for the plurality of SKUs may be determined, and if it is determined that an article having a particular SKU is likely to be scrapped, an action to prevent the article from being scraped may be performed.

3A and 3B illustrate an example method 300 for predicting and managing scrap likelihood associated with an inventory holding unit (SKU) using a trained predictive model. The method, or portions thereof, may be performed by the SCM system 117 . For example, a system may include one or more processors and memory that stores instructions that, when executed by one or more processors, cause the system to perform the steps illustrated in FIGS. 3A and 3B .

At step 302 , the one or more processors may be configured to receive a request to initiate a process associated with predicting scrap likelihood associated with a plurality of SKUs (ie, articles). For example, the SCM 117 may enable a user through interaction in a graphical user interface (GUI) of the user device to initiate a process for determining a value associated with predicting scrap likelihood for a plurality of SKUs. An indication may be received from In some embodiments, the request may originate from the SCM 117 and may be transmitted to a remote computer system to receive information associated with predicted scrap for a plurality of SKUs. Additionally or alternatively, the request or indication may be a selection of an element (eg, button, box, icon) on a graphical user interface of a computer system, such as SCM 117 , or a request from a first computer system (eg, user device). to a second computer system (eg, SCM 117 ).

At step 304 , the one or more processors may be configured to prepare input data associated with each SKU of the plurality of SKUs. Step 304 is further described below with respect to box 304A of FIG. 3C . At step 324 of FIG. 3C , the one or more processors may be configured to retrieve data from the database in response to receiving the request, the data comprising a plurality of scalable factors associated with each SKU of the plurality of SKUs. do. For example, source data including information associated with an SKU may be originally stored in a data warehouse. In response to the request, the one or more processors may unload the source data to storage and read the data into an application programming interface for extraction of necessary factors.

At step 326 , the one or more processors may be configured to scale a plurality of factors associated with each SKU of the plurality of SKUs. For example, prior to inputting data into the neural network, one or more processors may scale or normalize the input factors to transform the factors to a similar scale. Each factor may be associated with a different scale (eg, number of days, time, sales, occurrence), which may increase the difficulty of the problem being modeled. Thus, the input factors can be processed before being input to the neural network to scale or normalize the input factors to a common scale, such as a range of 0-1.

After step 326 of FIG. 3C , the process moves to step 306 of FIG. 3A . At step 306 , the one or more processors may be configured to input data associated with each SKU of the plurality of SKUs into the trained predictive model. For example, the one or more processors may input a plurality of scaled factors associated with each SKU among the plurality of SKUs into an open source library for machine learning. In some embodiments, the trained predictive model may be a deep neural network. In some embodiments, the trained predictive model may be a convolutional neural network. The trained neural network may include a first multi-dimensional input layer, a nested layer, and a one-dimensional output layer for outputting a single value.

At step 308 , the one or more processors may be configured to output a predictive value for each SKU of the plurality of SKUs from the trained predictive model. For example, after inputting the scaled factor to the open source library, the neural network may output a predicted value for each SKU among a plurality of SKUs. The predicted value may be a float number between 0 and 1, where a value of 1 may indicate scrap and a value of 0 may indicate non-scrap.

At step 310 , the one or more processors may be configured to measure the trained predictive model by calculating a set of values. In some embodiments, the set of values may include a percentage of gross merchandise value (GMV) lost corresponding to a recall value, an accuracy value, an F1 score, an F2 score, and a scrap reduction percentage. For example, one or more processors may perform back testing using a confusion matrix to calculate recall and accuracy values based on historical scrap data. Backtesting can be performed to check how well the neural network performed using the historical data stored in the database of the SCM 117, and the confusion matrix (i.e., the error matrix) is used to visualize the performance of the neural network based on the backtest. can be used to

By performing the backtest, one or more processors can be determined to be true positive (i.e., the SKU is correctly predicted as scrap), true negative (i.e. the SKU is correctly predicted as non-scrap), or false positive (i.e., the SKU is incorrectly predicted as scrap). A confusion matrix can be generated by counting the number of predicted) and false negative (ie, SKU incorrectly predicted as non-scrap) results. For example, one or more processors may backtest 1000 SKUs. By comparing the bag test results to historical data representing SKU articles that were actually scrapped, one or more processors determined that 450 out of 1000 SKUs correctly predicted as scrap, 350 correctly predicted as non-scrap, 150 incorrectly predicted as scrap, and 50 A dog can be determined to be incorrectly predicted as non-scrap. As a result of this, one or more processors may calculate values for recall and accuracy. For example, recall, which is the percentage of correct predictions among actual scrap cases, can be calculated by dividing the number of true positive results by the number of true scrap cases (ie true positive + false negative). In this case the recall would be 450/(450+50)=0.9. Accuracy, which is the percentage of correct predictions among scrap predictions, can be calculated by dividing the number of true positive results by the number of positive cases (ie true positive + false positive). In this case the accuracy would be 450/(450+150)=0.75. Based on the calculated recall and accuracy values, the one or more processors may calculate an F1 score and an F2 score. The F1 score represents a balance between recall and accuracy, with higher values when both recall and accuracy are at high levels. The F2 score is similar to the F1 score, but places more emphasis on recall than accuracy.

The one or more processors may be configured to match each SKU to four base cases of the confusion matrix to calculate a new GMV loss for each SKU of the plurality of SKUs. By calculating the new GMV loss value and calculating the difference between the old GMV loss value due to scrap and the new GMV loss value, the one or more processors may calculate a loss function representing the business effect of applying the prediction result. .

At step 312 , the one or more processors may be configured to generate a matrix including the calculated set of values. One such matrix is shown in FIG. 6 described below.

After step 312, the process 300 moves to step 314 of FIG. 3B. At step 314 , the one or more processors may be configured to select a first value of the set of calculated values corresponding to the threshold predicted values of the plurality of thresholds. In some embodiments, the one or more processors may be configured to select a scrap reduction percentage, wherein the scrap reduction percentage determines an optimal threshold value based on the selection process. The selection process may include a set of rules based on possible preventive actions that may be taken to improve the scrap reduction percentage. For example, if precautions are applied to a SKU, the scrap reduction percentage for that SKU may increase or remain the same. On the other hand, if precautions are not applied to a SKU, the scrap reduction percentage for that SKU may decrease or remain the same. Based on the set of rules, the one or more processors may determine and select a scrap reduction percentage, and may further determine a lowest GMV loss percentage associated with the scrap reduction percentage. For example, the one or more processors may select a scrap reduction percentage of 50% and determine that the lowest GMV loss percentage is -31.7% at a threshold of 0.5. In another example, the one or more processors may select a scrap reduction percentage of 40% and determine that the lowest GMV loss percentage is -38.6% at thresholds of 0.3, 0.4, and 0.5. To determine the most optimal threshold, the one or more processors may determine which recall and accuracy values are best balanced. For example, based on the matrix 600, at a threshold of 0.3, the recall is 0.62 and the accuracy is 0.70. At a threshold of 0.4, the recall is 0.60 and the accuracy is 0.71. At a threshold of 0.5, the recall is 0.59 and the accuracy is 0.73. Based on these values, one or more processors can determine that recall and accuracy are best balanced at a threshold of 0.4.

In some embodiments, the scrap reduction percentage is selected by a user using the user device to communicate with a computer system (eg, SCM 117 ) via a graphical user interface (GUI). For example, the one or more processors may receive, via interaction on the graphical user interface, an instruction from the user to select a desired average scrap quantity reduction percentage based on a set of rules. Based on the selected percentage, the one or more processors may determine a lowest GMV loss percentage of the plurality of GMV loss percentages, and may further determine a threshold value corresponding to the GMV loss percentage. Based on the determination of the threshold, the one or more processors may set a threshold prediction and determine which SKU articles are likely to be scraped based on the value output by the predictive model.

At step 316 , the one or more processors may be configured to compare a prediction output from the trained prediction model with a threshold prediction for each SKU of the plurality of SKUs. For example, based on the comparison, the one or more processors may determine that the prediction is greater than or less than a threshold prediction.

At step 318 , the one or more processors may be configured to determine a scrapability for each SKU of the plurality of SKUs based on the comparison. For example, an SKU article would be predicted to be likely to be scraped based on a determination that the predicted value associated with the SKU is greater than a threshold predicted value. On the other hand, based on a determination that the predicted value associated with the SKU is less than the threshold predicted value, the SKU article will be predicted as unlikely to be scrapped.

At step 320 , the one or more processors may be configured to perform at least one of a plurality of preventive actions based on a scrap likelihood indicating that scrap is likely. For example, the one or more processors may update the website to lower the price associated with the SKU, update the order quantity for the SKU in the management database, or remove the SKU from the list of SKUs flagged as out of stock on the website. can do. In another example, the one or more processors may update the website home page to display items associated with the SKU or generate instructions for transmission from the first computer system to a second computer system associated with the recommended action. In some embodiments, the one or more processors display a message on the device of the SCM 117 or send a notification to the user using the user device of the SCM 117 to warn the user to order fewer SKU items or to add to the SKU items. can offer to sell. In some embodiments, the one or more processors may be configured to automatically perform at least one of a plurality of preventive actions in response to a scrap likelihood indicating that scrap is likely. In some embodiments, the one or more processors may wait to determine a scrapability for each SKU of all SKUs before performing the preventive action. Alternatively, the one or more processors may take preventive action as soon as the SKU is determined to be predicted scrap.

At step 322 , the one or more processors may be configured to take no preventive action based on a scrap probability indicating that it is unlikely to be scrapped. For example, after determining that the SKU is unlikely to be scrapped, the process may return to step 318 to determine a scrap probability for a next SKU of the plurality of SKUs.

4 shows an exemplary block diagram 400 illustrating the flow of preparing input data for a trained neural network. Source data including information associated with the SKU is stored in a data warehouse 410 such as Amazon Redshift. Information related to SKUs includes: shelf life, average sales, sales trends, inventory, out of stock (OOS) time, code life, available days, backlog, last sales, last scrap, last order quantity, minimum order quantity (MOQ), etc. It may include the same factor, but is not limited thereto. One or more processors access source data from data warehouse 410 using a structural query language (SQL) to unload the data to storage 420, such as storage service Amazon S3, and PySpark (i.e., Python's Data may be read into an application programming interface 430 such as an interface for Apache Spark. The application programming interface 430 may use an open source tool such as ELI5 (ie, a Python library) to extract necessary factors from data to analyze factor importance. When the factors associated with each SKU among the plurality of SKUs are prepared, one or more processors may input the prepared data into an open source library 440 such as Tensorflow for machine learning.

5, described below, depicts an example diagram 500 illustrating the flow of information through a predictive model 510 . For example, the one or more processors may have data associated with 100 SKUs comprising several factors including inventory backlog, sales trends, over-prediction, OOS time, code life sellable days, average sales, MOQ, shelf life, and inventory. 502 may be input to predictive model 510 . After the data 502 is input to the predictive model 510 , the predictive model 510 may output data 504 including a prediction value for each SKU out of 100 SKUs. The predicted value of the data 504 output by the predictive model 510 may be a floating point number from 0 to 1, where a value of 1 indicates scrap and a value of 0 indicates non-scrap.

6 shows an exemplary generated matrix 600 including a set of computed values. Matrix 600 includes variable thresholds and average scrap quantity reduction percentages. Each average percentage of scrap quantity to be reduced has a range of GMV loss percentages corresponding to different thresholds. For example, if the scrap reduction percentage is 50%, there is a range of GMV loss percentages including the highest percentage GMV loss is 9.61% and the lowest percentage GMV loss is -31.7%. Matrix 600 also includes values for recall and accuracy at different thresholds, where the thresholds are threshold predictions associated with the predictions output by predictive model 510 .

The GMV loss percentage corresponds to a threshold value, which may be used to set a threshold value for determining which prediction value output by the prediction model 510 can be predicted as scrap. If the threshold is set to a larger number (i.e. greater than 0.5) there will be fewer instances predicted to be scrap, and if the threshold is set to a smaller number (i.e. less than 0.5), more instances will be predicted to be scrap . In some embodiments, matrix 600 may also include values for F1 scores and F2 scores. Additionally or alternatively, matrix 600 may be displayed for viewing by a user on a user device.

In some embodiments, the scrap reduction percentage is selected by a user using the user device to communicate with a computer system (eg, SCM 117 ) via a graphical user interface (GUI). For example, the one or more processors select, from the user through interaction on the graphical user interface, a desired 50% average scrap quantity reduction percentage based on a calculated value (eg, recall, accuracy) represented in matrix 600 . instructions can be received. Based on the selected percentage of 50%, the one or more processors may determine a lowest percentage of loss of GMV of -31.7% of the plurality of percentages of loss of GMV, and may further determine a threshold value of 0.5 corresponding to the percentage of loss of GMV. Based on the determination of the threshold value, SCM 117 may set the threshold prediction value to 0.5, and may determine which SKU articles are likely to be scraped based on the value output by predictive model 510 . For example, as shown in data 504 ( FIG. 5 ), by comparing the predicted value of 0.61 for SKU 1 with a threshold set to 0.5, the one or more processors can determine that the predicted value is greater than the threshold predicted value. Meanwhile, by comparing the predicted value of 0.32 for SKU 2 with a threshold value set to 0.5, the one or more processors may determine that the predicted value is less than the threshold prediction value. An article with SKU 1 will be predicted as likely to be scraped based on a determination that the prediction value associated with SKU 1 is greater than the threshold prediction value. On the other hand, based on a determination that the predicted value associated with SKU 2 is less than the threshold predicted value, the article associated with SKU 2 will be predicted as unlikely to be scrapped. Based on the scrap probability indicating that it is likely to be scraped, the one or more processors may take preventive action as discussed above. On the other hand, one or more processors may not take a preventive action based on a scrap probability indicating that it is unlikely to be scrapped. For example, after determining that an article with SKU 2 is unlikely to be scrapped, the process may return to step 318 to determine a scrap likelihood for SKU 3 .

While the present disclosure has been shown and described with reference to specific embodiments thereof, it will be understood that the present disclosure may be practiced without modification in other environments. The foregoing description has been presented for purposes of illustration. It is not exhaustive, nor is it limited to the precise form or embodiment disclosed. Modifications and adaptations will become apparent to those skilled in the art from consideration of the specification and practice of the disclosed embodiments. Additionally, although aspects of the disclosed embodiments are described as being stored in memory, those of ordinary skill in the art will recognize that these aspects may also be implemented in a second storage device, such as a hard disk or CD ROM or other form of RAM or ROM; It will be understood that the storage may be stored on other types of computer readable media such as USB media, DVD, Blu-ray or other optical drive media.

A computer program based on the described description and disclosed method is within the skill of the skilled developer. The various programs or program modules may be created using any techniques known to those of ordinary skill in the art or may be designed in conjunction with existing software. For example, a program section or program module can contain .Net Framework, .Net Compact Framework (and related languages such as Visual Basic, C, etc.), Java, C++, Objective-C, HTML, HTML/AJAX combination, XML, or Java applet. (applet) can be designed in or by embedded HTML.

Moreover, although exemplary embodiments have been described herein, the scope of any and all embodiments having equivalent elements, modifications, omissions, combinations, adaptations, and/or variations (eg, in aspects across various embodiments) is scoped. will be recognized by those skilled in the art based on the present disclosure. Limitations in the claims are to be construed broadly based on the language used in the claims, and are not limited to the examples set forth herein or to the pending application. Examples shall be construed as non-exclusive. Furthermore, the steps of the disclosed method may be modified in any way, including reordering the steps and/or inserting or deleting steps. Therefore, the specification and examples are intended to be regarded as illustrative only, with the true scope and spirit being indicated by the full scope of the following claims and their equivalents.

Claims (20)

A computer-implemented system for predicting and managing predicted scrap items, the computer-implemented system comprising:
one or more memory devices to store instructions;
one or more processors configured to execute the instructions to perform an operation, the operation comprising:
receiving a request to initiate a process associated with predicting scrap likelihood associated with a plurality of Stock Keeping Units (SKUs);
in response to receiving the request, preparing input data associated with each SKU of the plurality of SKUs;
inputting the input data associated with each SKU of the plurality of SKUs into a trained prediction model;
outputting a prediction value for each SKU among the plurality of SKUs from the trained prediction model;
measuring the trained predictive model by calculating a set of values; and
preparing information associated with the request by generating a matrix comprising at least a subset of the set of computed values;
selecting a first value of the set of calculated values corresponding to a threshold predicted value of a plurality of thresholds;
comparing the prediction value output from the trained prediction model with the threshold prediction value for each SKU of the plurality of SKUs;
determining a scrap probability for each SKU of the plurality of SKUs based on the comparison; and
and performing at least one of a plurality of preventive actions based on the scrap likelihood indicating a scrap potential. According to claim 1,
wherein the predictive model comprises a deep neural network. According to claim 1,
Preparing input data associated with each SKU among the plurality of SKUs includes:
in response to receiving the request, retrieving data from a database, the data comprising a plurality of scalable factors associated with each SKU of the plurality of SKUs; and
and scaling the plurality of factors associated with each SKU of the plurality of SKUs. 4. The method of claim 3,
Entering the input data associated with each SKU of the plurality of SKUs into a trained predictive model includes inputting the plurality of scaled factors associated with each SKU of the plurality of SKUs into the trained predictive model. which is a computer implemented system. According to claim 1,
Performing at least one of a plurality of preventive measures may include updating the website to lower the price associated with the SKU, updating the order quantity for the SKU in a management database, or flagging the SKU as out of stock on the website. ) removing the SKU from the SKU list. According to claim 1,
and the processor is further configured to execute the instructions to perform an action comprising not performing a preventive action based on the likelihood of scrap indicating that it is unlikely to be scrapped. According to claim 1,
wherein the set of calculated values comprises a recall value, a precision value, an F1 score, an F2 score, and a percentage loss of gross merchandise value (GMV). According to claim 1,
Determining a scrap probability for each SKU of the plurality of SKUs based on the comparison includes:
when the predicted value is greater than the threshold predicted value, determining that one or more articles having a particular SKU are likely to be scrapped; and
and when the predicted value is less than the threshold predicted value, determining that one or more articles having a particular SKU are unlikely to be scrapped. According to claim 1,
The one or more processors are further configured to execute the instructions to perform an operation, the operation comprising:
determining the corresponding threshold predicted value of the plurality of threshold values based on a GMV loss percentage of a plurality of GMV loss percentages and the selected first value, wherein the selected first value is a selected scrap quantity reduction percentage. comprising a computer-implemented system. According to claim 1,
and measuring the trained predictive model by calculating a set of values comprises using a confusion matrix. A computer-implemented method for predicting and managing predicted scrap items, the computer-implemented method comprising:
receiving a request to initiate a process associated with predicting scrap likelihood associated with a plurality of inventory holding units (SKUs);
in response to receiving the request, preparing input data associated with each SKU of the plurality of SKUs;
inputting the input data associated with each SKU of the plurality of SKUs into a trained predictive model;
outputting a prediction value for each SKU among the plurality of SKUs from the trained prediction model;
measuring the trained predictive model by calculating a set of values; and
preparing information associated with the request by generating a matrix comprising at least a subset of the set of computed values;
selecting a first value of the set of calculated values corresponding to a threshold predicted value of a plurality of thresholds;
comparing the prediction value output from the trained prediction model with the threshold prediction value for each SKU of the plurality of SKUs;
determining a scrap probability for each SKU of the plurality of SKUs based on the comparison; and
and performing at least one of a plurality of preventive actions based on the likelihood of scrap indicating a likelihood of being scrapped. 12. The method of claim 11,
wherein the predictive model comprises a deep neural network. 12. The method of claim 11,
The step of preparing input data associated with each SKU among the plurality of SKUs includes:
in response to receiving the request, retrieving data from a database, the data comprising a plurality of scalable factors associated with each SKU of the plurality of SKUs; and
and scaling the plurality of factors associated with each SKU of the plurality of SKUs. 14. The method of claim 13,
The inputting of the input data associated with each SKU of the plurality of SKUs into a trained predictive model may include: inputting the plurality of scaled factors associated with each SKU of the plurality of SKUs into the trained predictive model; A computer-implemented method comprising: 12. The method of claim 11,
Performing at least one of a plurality of preventive actions may include updating a website to lower a price associated with the SKU, updating an order quantity for the SKU in a management database, or an SKU flagged as out of stock on the website. and removing the SKU from a list. 12. The method of claim 11,
The method further comprising not performing a preventive action based on the likelihood of scrap indicating that it is unlikely to be scrapped. 12. The method of claim 11,
wherein the set of calculated values includes a recall value, an accuracy value, an F1 score, an F2 score, and a percentage of gross merchandise value (GMV) loss. 12. The method of claim 11,
Determining a scrap probability for each SKU of the plurality of SKUs based on the comparison includes:
when the predicted value is greater than the threshold predicted value, determining that one or more articles having a particular SKU are likely to be scrapped; and
when the predicted value is less than the threshold predicted value, determining that one or more articles with a particular SKU are unlikely to be scrapped. 12. The method of claim 11,
The method is:
determining the corresponding threshold predicted value of the plurality of threshold values based on a GMV loss percentage of a plurality of GMV loss percentages and the selected first value, wherein the selected first value is a selected scrap quantity reduction percentage. further comprising a computer implemented method. A computer-implemented system for predicting and managing predicted scrap items, the computer-implemented system comprising:
one or more memory devices to store instructions;
one or more processors configured to execute the instructions to perform an operation, the operation comprising:
receiving a request to initiate a process associated with predicting scrap likelihood associated with a plurality of inventory holding units (SKUs);
in response to receiving the request, preparing input data associated with each SKU of the plurality of SKUs;
inputting the input data associated with each SKU of the plurality of SKUs into a trained neural network;
outputting a predicted value for each SKU among the plurality of SKUs from the trained neural network;
measuring the trained neural network by calculating a set of values; and
preparing information associated with the request by generating a matrix comprising at least a subset of the set of computed values:
selecting a first value of the set of computed values corresponding to a threshold predicted value of a plurality of thresholds;
comparing the prediction value output from the trained neural network with the threshold prediction value for each SKU of the plurality of SKUs;
determining a scrap probability for each SKU of the plurality of SKUs based on the comparison; and
and performing at least one of a plurality of preventive actions based on the scrap likelihood indicating a scrap potential.

Images